SAD
Summary
Facts
Influence
Introspection
Stages
Self Recognition
ID Leverage
Anti Imitation
Summary
SAD Score on selected models
Base vs Chat Models
Effect of SP by Model
Effect of SP by Task
Correlations between SAD and MMLU
Correlations between SAD, SAD Categories and MMLU
Facts
Facts Category
Human Defaults
LLMs
Which LLM
Names
Influence
Influence Category
Introspection
Introspection Category
Count Tokens
Predict Words
Rules
Stages
Stages Category
Stages Full
Stages Oversight
Self Recognition
Self Recognition Category
Who
Groups
ID Leverage
ID Leverage Category
Entity Name
Multihop
Anti Imitation
Anti Imitation Category
Output Control
Do Not Imitate